In case anyone's interested (and sorry if this was spam for you, I used a
clear subject line so hopefully you can just ignore it if it is!), below my
signature are details for this talk I'm giving:
Tomorrow at 11 am EST (Fri 22, Jan): Joseph will be giving a Vector Talk on
Adapting Real-World Experimentation To Balance Enhancement of User
Experiences with Statistically Robust Scientific Discovery.
Click this link to add to Google Calendar
<https://www.google.com/calendar/render?action=TEMPLATE&text=Vector%20Instit…>
with
zoom link and abstract.
Or email angelina.liu(a)mail.utoronto.ca and cc williams(a)cs.toronto.edu for a
copy of the recording.
Here our lab's "HCI talk"
http://www.josephjaywilliams.com/prospectivestudents#TOC-HCI-Human-Computer…
But the talk will be pretty accessible with zero background in
statistics/machine learning. The key ideas are: If you run randomized
experiments in the real world, how can you make them adaptive experiments?
By using machine learning to rapidly use data to give people the better
interventions, while also enabling reliable statistical analysis of the
data?
Actually, it will be great to get feedback on this talk from HCI people,
like whether we are convincing you to actually use these methods for
randomized experiments you would run.
Anyone from my lab (or DGP) who joins the talk, please let me know if you
are willing to use the zoom 'thumbs up' emoticon and just flash that EVERY
TWO MINUTES! Even if you think my speed is fine, flash it :D. That will
greatly increase the audience's comprehension and user experience!
Joseph
*INFORMATION ON TALK:*
Subject Line: Fri 22 Jan 11 am: Vector Talk on Adapting Real-World
Experimentation to Balance Enhancement of User Experiences with
Statistically Robust Scientific Discovery
Joseph Jay Wiliams <http://www.josephjaywilliams.com/> is giving a talk in
the Vector Institute for AI seminar this Friday: Jan 22nd at 11am EST. (Click
to add to Calendar
<https://www.google.com/calendar/render?action=TEMPLATE&text=Vector%20Instit…>).
To
request a recording & slides, email angelina.liu(a)mail.utoronto.ca and cc
williams(a)cs.toronto.edu.
Link to register
<https://vectorinstitute.zoom.us/meeting/register/tJ0qde2uqjwoGdEcu1GeiqDS3a…>
Meeting ID: 997 2464 7235
Password: 123456
Zoom Meeting Link
<https://vectorinstitute.zoom.us/w/99724647235?tk=ljUeWXxozZn8COEiN82E9F0TnJ…>
Short Title:Adapting Real-World Experimentation To Balance Enhancement of
User Experiences with Statistically Robust Scientific Discovery
Long Title:Perpetually Enhancing User Interfaces in Tandem with Advancing
Scientific Research in Education & Mental Health: Enabling Reliable
Statistical Analysis of the Data Collected by Algorithms that Trade Off
Exploration & Exploitation
How can we transform the everyday technology people use into intelligent,
self-improving systems? For example, how can we perpetually enhance text
messages for managing stress, or personalize explanations in online
courses? Our work explores the use of randomizedadaptiveexperiments that
test alternative actions (e.g. text messages, explanations), aiming to gain
greater statistical confidence about the value of actions, in tandem with
rapidly using this data to give better actions to future users.
To help characterize the problems that arise in statistical analysis of
data collected while trading off exploration and exploitation, we present a
real-world case study of applying the multi-armed bandit algorithm TS
(Thompson Sampling) to adaptive experiments. TS aims to assign people to
actions in proportion to the probability those actions are optimal. We
present empirical results on how the reliability of statistical analysis is
impacted by Thompson Sampling, compared to a traditional experiment using
uniform random assignment. This helps characterize a substantial problem to
be solved – using a reward maximizing algorithm can cause substantial
issues in statistical analysis of the data. More precisely, an adaptive
algorithm can increase both false positives (believing actions have
different effects when they do not) and false negatives (failing to detect
differences between actions). We show how statistical analyses can be
modified to take into account properties of the algorithm, but that these
do not fully address the problem raised.
We therefore introduce an algorithm which assigns a proportion of
participants uniformly randomly and the remaining participants via Thompson
sampling. The probability that a participant is assigned using Uniform
Random (UR) allocation is set to the posterior probability that the
difference between two arms is 'small' (below a certain threshold),
allowing for more UR exploration when there is little or no reward to be
gained by exploiting. The resulting data can enable more accurate
statistical inferences from hypothesis testing by detecting small effects
when they exist (reducing false negatives) and reducing false positives.
The work we present aims to surface the underappreciated complexity of
using adaptive experimentation to both enable scientific/statistical
discovery and help real-world users. The current work takes a first step
towards computationally characterizing some of the problems that arise, and
what potential solutions might look like, in order to inform and invite
multidisciplinary collaboration between researchers in machine learning,
statistics, and the social-behavioral sciences.
Bio:Joseph Jay Williams is an Assistant Professor in Computer Science (and
a Vector Institute Faculty Affiliate, with courtesy appointments in
Statistics & Psychology) at the University of Toronto, leading the
Intelligent Adaptive Interventions research group. He was previously an
Assistant Professor at the National University of Singapore's School of
Computing in the department of Information Systems & Analytics, a Research
Fellow at Harvard's Office of the Vice Provost for Advances in Learning,
and a member of the Intelligent Interactive Systems Group in Computer
Science. He completed a postdoc at Stanford University in Summer 2014,
working with the Office of the Vice Provost for Online Learning and the
Open Learning Initiative. He received his PhD from UC Berkeley in
Computational Cognitive Science (with Tom Griffiths and Tania Lombrozo),
where he applied Bayesian statistics and machine learning to model how
people learn and reason. He received his B.Sc. from University of Toronto
in Cognitive Science, Artificial Intelligence and Mathematics, and is
originally from Trinidad and Tobago. More information about the Intelligent
Adaptive Intervention group's research and papers is at
www.josephjaywilliams.com <http://www.josephjaywilliams.com/>.
Joseph
Joseph Jay Williams
www.josephjaywilliams.com
Assistant Professor
Department of Computer Science, University of Toronto
Intelligent Adaptive Interventions (IAI) research group
Hi DGP!
Some of us have talked about starting a discussion or reading group on Health + CS related topics. We have a slack channel #health that we’ve started and we’ve been thinking about making it a weekly session to discuss over zoom as well.
The meetings will be Wednesdays and will start this week, January 13, 2021 at 12pm ET. The zoom link is: https://utoronto.zoom.us/j/7301759089. We will be going over some of the topics we want to discuss and talk about the structure of how the meetings will proceed.
If you’re interested or want to know more, please join the slack channel and come to the meeting!
Hope to see many of your there!
Best,
Brenna